Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

1.

3 Modeling Ultrasonic Phased Array Systems 9


Vmnr ( f ) = tmr ( f ) FmnB ( f ) (1.2)

This receiving transfer function is a function of the electrical impedance and gain
present in the receiving circuits, the wiring/cabling present, and the electrical imped-
ance and sensitivity of the mth piezoelectric element [Schmerr-Song]. The blocked
force appearing in Eq. (1.2) is defined as the force exerted on the face of the receiv-
ing element when the face of that element is held rigidly fixed. In [Schmerr-Song]
and in Chap. 9 it is shown how this blocked force arises naturally in describing the
sound reception process for an ultrasonic system.
From Eqs. (1.1) and (1.2) we see that in an ultrasonic measurement process in-
volving a pair of elements, where the waves are generated by the nth element and
received by the mth element, the received voltages, Vmnr, are given by

Vmnr ( f ) FmnB ( f ) Fnt ( f ) i
Vmnr ( f ) = Vn ( f )
FmnB ( f ) Fnt ( f ) Vmi ( f ) (1.3)
= tmr ( f )tng ( f )tmn
a
( f )Vni ( f )
= smn ( f )tmn ( f )
a

where the acoustic/elastic transfer functions, tmn a


( f ), are defined as tmn
a
≡ FmnB / Fnt
and the system functions, smn ( f ) , are given by
(1.4)
smn ( f ) = tmr ( f )tng ( f )Vni ( f )

( )
If we apply time delays, ∆tn , ∆tm on sound generation and reception, respectively,
g r

and also apodization weights (C ng, C mr ), on sound generation and reception, respec-
tively, then in the frequency domain these effects can easily be incorporated into
Eq. (1.3) (see Chaps. 4 and 7) as

( ) ( )
Vmnr ( f ) = C ng C mr exp 2πif ∆tng exp 2πif ∆tmr smn ( f )tmn
a
( f ). (1.5)

Equation (1.5) is a very general model for an ultrasonic phased array measurement
system as it describes the voltages received by all possible pairs of sending and
receiving elements. This equation also shows that all the electrical and electrome-
chanical parts of the system response can be characterized by the system functions,
smn ( f ) and all the wave propagation and scattering processes present can be char-
acterized by the acoustic/elastic transfer functions, tmn
a
( f ).
In Chap. 10 it will be shown that the system functions can be obtained experi-
mentally by measuring the received voltage, Vmnr ( f ), for various element pairs in a
reference configuration where the acoustic/elastic transfer functions, tmn a
( f ) are
known. These acoustic/elastic transfer functions will also be derived in Chap. 10 for
a convenient calibration setup. Then from Eq. (1.3) we have formally
 Vmnr ( f )
smn ( f ) = a
. (1.6)
tmn (f)
10 1 Introduction

In practice, the direct division in Eq. (1.6) is replaced by a Wiener filter to desensi-


tize this deconvolution process to noise [Schmerr-Song]. If M = N so there are M
transmitting and receiving elements, then a total of M 2 different system functions
are needed to characterize the behavior of the entire array. Even if one assumes that
the system functions and acoustic/elastic transfer functions are symmetric, one still
has a total of M ( M +1) / 2 different system functions that would have to be ob-
tained from Eq. (1.6). For a 16 element linear array, for example, this would corre-
spond to 136 different calibration experiments that would be needed to characterize
the entire array. Fortunately, however, phased arrays are normally made with nomi-
nally identical driving and receiving circuits, wiring, and piezoelectric elements, so
it is not surprising, as found in Chap. 10, that the measured system functions may
also be nominally identical, i.e. smn ( f ) ≅ s ( f ). In this case only one calibration
experiment is needed to obtain this system function and Eq. (1.3) becomes

(1.7)
Vmnr ( f ) = s ( f )tmn
a
(f)

and the general model of a phased array system with time delay law and apodization
laws, Eq. (1.5) becomes

( ) (
Vmnr ( f ) = C ng C mr exp 2 if ∆tng exp 2 if ∆tmr s ( f )tmn
(1.8) a
)
( f ).

Equation (1.8) shows that if the system function is determined experimentally and
the time delays and apodization weights specified, knowledge of all the acoustic/
elastic transfer functions present is required to simulate the received voltage. These
transfer functions are functions of the waves generated by the driving elements and
propagating in the media present, the waves scattered by a reflector or flaw present,
and the waves propagated to the receiving elements. It is not possible to determine
these propagating and scattered waves experimentally so that it is necessary to have
explicit models of these wave processes and how they contribute to the acoustic/
elastic transfer functions.
In order to describe the ultrasonic wave fields generated by the driving elements
in the acoustic/elastic transfer function it is necessary to develop an appropriate
mathematical model of the array as a set of acoustical sources. In the frequency
domain these field quantities are all functions of the spatial variables ( x, y, z )
and the frequency, f. For a large, single element transducer such as the one shown
in Fig. 1.11a a simple but effective frequency domain model that has been used
assumes that the transducer acts like a velocity source with the normal velocity,
vz ( x, y, z = 0, f ), on the face of the transducer given by

 v0 ( f ) on S
vz ( x, y, z = 0, f ) =  , (1.9)
0 otherwise

where S is the active area of the transducer face (see Fig. 1.11a). This model as-
sumes that the entire face of the transducer acts in unison in a piston-like manner,
i.e. it is a piston transducer model. Piston behavior has been successfully used to
1.3 Modeling Ultrasonic Phased Array Systems 11

Fig. 1.11   a A large, single


element rectangular trans-
ducer, and b its representation
as a piston velocity source
acting over the area S that is
embedded in an infinite, rigid
baffle

Fig. 1.12   a A linear phased


array and b a model of an
individual element as a piston
velocity source embedded in
a rigid baffle

model many large single element transducers [Schmerr], [Schmerr-Song]. In our


transducer array modeling we will also assume that each element of the array also
acts like a piston. Equation (1.9) shows that in this piston transducer model the
piston source is embedded in an infinite planar rigid baffle, as shown in Fig. 1.11b.
It is possible to replace the actual 3-D geometry of the transducer in Fig. 1.11a
by a piston source acting in a planar rigid baffle since for large single element
transducers the transducer crystal is supported along its edges by a relatively rigid
case. Also, as we will see, for example in Chaps. 3 and 6, for large, single element
transducers, which typically operate at MHz frequencies in NDE tests, the beam
of sound generated by the transducer is well-collimated, i.e. it is significant only
in a relatively small region directly ahead of the transducer. Thus, the fields on the
plane z = 0 outside the region S are typically very small anyway, so that the rigid
baffle assumption does not affect the fields significantly. For phased arrays such
as the linear and 2-D arrays shown in Figs. 1.12a and 1.13a, however, the phased
array elements are separated by gaps filled with an epoxy-like material (which is
12 1 Introduction

Fig. 1.13   a A 2-D phased


array and b a model of an
individual element as a piston
velocity source embedded in
a rigid baffle

not as rigid as the elements themselves) and at MHz frequencies each element of
the array does not generate a well-collimated beam of sound in all directions so one
should examine the validity of applying the rigid baffle model of Eq. (1.9) to each
element of the array. In modeling phased arrays in this book we will still assume
that each element is surrounded by a rigid baffle, as shown in Figs. 1.12b and 1.13b
but in Chap. 14 we will model the radiation of an element when it is surrounded by
a finite acoustic impedance baffle and discuss how to determine experimentally if
such effects are important. Other modeling issues associated with arrays will also
be discussed in Chap. 14.
The voltages, Vmnr ( f ) (m = 1, 2,...M ),(n = 1, 2,...N ) , are all the possible measure-
ments that can be made with a phased array system having M sending elements and
N receiving elements. If the data from the measurement of all these voltages sepa-
rately is available, the measurement is said to be one of full matrix capture (FMC).
Full matrix capture provides the largest amount of information that is available
from a phased array system and is shown in Fig. 1.14a as a fully filled information
matrix of senders and receivers. If only the same individual element is used as both
a sender and receiver, then the collection of these pulse-echo responses is shown in
the information matrix of Fig. 1.14b. In Chap. 13 we will develop imaging methods
based on both full-matrix capture as well as the pulse-echo responses. In develop-
ing the FMC case, we will also need to consider the case of a single sending ele-
ment and multiple receiving elements, as shown in Fig. 1.14c. This case is known
in the seismology literature as a common shot response. Of course, many other
combinations of the sending and receiving elements can be used for measurements
and imaging, but the FMC and pulse-echo cases are those most commonly found
in practice.
In an ultrasonic measurement, if the responses of all M sending elements and
N receiving elements are measured then we can simply sum all these responses to
obtain a single response, V r ( f ) , given by
 M N

( ) ( )
V r ( f ) = ∑ ∑ C ng C mr exp 2πif ∆tng exp 2πif ∆tmr smn ( f )tma n ( f ). (1.10)
m =1 n =1
1.4 Book Outline 13

Fig. 1.14   Information


matrices showing the data
available with combinations
of sending and receiving
elements for a full matrix
capture (all element pairs), b
pulse-echo responses (same
sending/receiving pairs), and
c a single sending element
used with all the receivers

This measured voltage is then analogous to what would be measured by a single ele-
ment transducer of a size comparable to the whole array, but where the array trans-
ducer beam can be tailored by the steering, focusing and apodization terms. Com-
mercial phased array systems typically provide this summed signal as an output, as
well as standard images formed with the array signals such as B-scans, S-scans, etc.
However, with full matrix capture capabilities, a phased array system allows the
user to manipulate the array data and form images in ways that are not possible with
the output signal of Eq. (1.10).

1.4 Book Outline

This book is divided into essentially three sections. The first section, covering
Chaps. 2–5, idealizes arrays as 1-D elements radiating waves in two dimensions.
This assumption allows us to discuss many modeling issues and important concepts
such as beam steering, focusing, and the existence of grating lobes in a very simple
framework. This section also provides an ideal source of materials for introducing
students to phased arrays and describes some MATLAB® functions and scripts that
can be used to simulate the behavior of a phased array.
The second section of the book, in Chaps. 6–11, develops a complete model
of a phased array ultrasonic measurement system. Phased array beam models are
developed in detail in Chaps. 6 and 7 and the time delay laws that can be used
to control the behavior of an array are obtained in Chap. 8. A complete linear
systems model of an ultrasonic phased array measurement system is developed
in Chap. 9 where the system response is divided into a system function that de-
scribes the electrical and electro-mechanical parts of the system associated with
a sending and receiving pair of elements, and an acoustic/elastic transfer func-
tion that describes all the acoustic and elastic wave propagation and scattering
fields present between those sending and receiving elements, as discussed earlier
in this Chapter. Chapter 10 shows how the system function for each element can
14 1 Introduction

be measured experimentally in a calibration setup while Chap. 11 uses reciproc-


ity relations to obtain an expression for the acoustic/elastic transfer function for a
flaw measurement system in terms of the incident and scattered waves at the flaw
surface in a form similar to that originally developed by Auld [7]. The combina-
tion of the system function and the acoustic/elastic transfer function then gives an
explicit expression for the measured voltage from each pair of sending/receiving
elements. This expression is called an ultrasonic measurement model. It is also
shown in Chap. 11 how for small flaws this general measurement model can be
reduced to a Thompson-Gray type of form [8] where the flaw response is obtained
as an explicit and separate part of the overall expression for the received voltage.
Examples are given of how this reduced measurement model can be used to predict
the measured response of some simple reflectors.
Since ultrasonic phased array flaw measurement systems are commonly used to
generate images of the flaws present, the third section of the book, Chaps. 12 and 13,
are devoted to the fundamentals of imaging. In Chap. 12, two commonly used ad-
hoc imaging methods, the Synthetic Aperture Focusing Technique (SAFT) and the
Total Focusing Method (TFM) are first discussed. Then it is shown how, for simple
2-D problems, Thompson-Gray measurement models can be inverted to produce an
explicit image of the surface geometry and reflectivity of a flaw in a form called
an imaging measurement model. These imaging measurement models are shown
to be closely related to the Physical Optics Far Field Inverse Scattering (POFFIS)
method and also to SAFT and TFM imaging. The nature of the images generated
with imaging measurement models are also described in Chap. 12 through a number
of “exact” simulations. In Chap. 13 imaging measurement models are more fully
developed for both large and small flaws in 3-D, leading to a unified framework
of imaging that generalizes SAFT, TFM, and POFFIS imaging and rationally de-
scribes the terms inherently present in the imaging process. The implications that
these imaging measurement models have on quantitative flaw characterization are
also discussed.
Finally, in Chap. 14, some of the explicit assumptions used in the development of
the array beam models used in the previous Chapters are re-examined. Specifically,
as discussed previously in Sect. 1.3, the assumption that an array element acts as a
velocity source in a surrounding rigid baffle is relaxed and a more general model is
developed where the baffle is allowed to have finite acoustic impedance.
There are also three Appendices. Appendices A and B provide detailed deriva-
tions of several factors that appear in the development of imaging measurement
models. Appendix C gives complete Code Listings for the MATLAB® functions
and scripts described in the book.
References [1] and [2] will be referred to often in this book and are listed as
[Schmerr] and [Schmerr-Song], respectively, in this and later Chapters.
References 15

References

1. L.W. Schmerr, Fundamentals of Ultrasonic Nondestructive Evaluation—A Modeling Approach


(Plenum Press, New York, 1998)
2. L.W. Schmerr, S.-J. Song, Ultrasonic Nondestructive Evaluation Systems—Models and Mea-
surements (Springer, New York, 2007)
3. T.L. Szabo,, Diagnostic Ultrasound: Inside Out (Academic Press, New York, 2004)
4. K.K. Shung, Diagnostic Ultrasound: Imaging and Blood Flow Measurements (CRC Press,
Boca Raton, 2006)
5. P.N.T. Wells, Ultrasonic imaging of the human body. Rep. Prog. Phys. 62, 671–722 (1999)
6. N. Dube, Introduction to Phased Array Technology Applications. R/D Tech. (2004). (available
from www.olympus-ims.com Advanced NDT Series Books)
7. B.A. Auld, General electromechanical reciprocity relations applied to the calculation of elastic
wave scattering coefficients. Wave Motion. 1, 3–10 (1979)
8. R.B. Thompson, T. A. Gray, A model relating ultrasonic scattering measurements through
liquid-solid interfaces to unbounded medium scattering amplitudes. J. Acoust. Soc. Am. 74,
140–146 (1983)
Chapter 2
Acoustic Field of a 1-D Array Element

As discussed in Chap. 1, an ultrasonic phased array is composed of many small


acoustic sending and receiving elements, each of which acts as an individual send-
ing or receiving transducer. In this Chapter we will develop models of the acoustic
waves generated by a single element and describe how the nature of this wave
field depends on the size of the element and its motion. Models that simulate the
radiation of a single array element will be generated explicitly in MATLAB®. The
superposition of a number of these single element models with different driving
excitations will then give us a complete model of a multi-element phased array
transducer, as shown in Chap. 4. To keep the discussion as simple as possible in this
Chapter the single element will be treated as a 1-D source of sound radiating two-
dimensional waves into a fluid or through a planar interface between two fluids.
Although both linear and 2-D arrays are composed of 2-D elements which produce
sound waves traveling in three dimensions, the physics of wave propagation is simi-
lar for both 1-D and 2-D elements so that we can learn much of the fundamentals of
sound generation with these simplified models. In Chaps. 6 and 7 we will discuss
the corresponding three-dimensional models and wave fields of single elements and
phased arrays.

2.1  Single Element Transducer Models (2-D)

The basic setup we will use to describe a single element transducer is shown in
Fig. 2.1. We will treat the transducer as a velocity source located on the plane z = 0
where a normal velocity, vz ( x, t ) , as a function of the location, x, and time, t, is
generated over a finite length [− b, b] in the x-direction and [ −∞, +∞ ] in the y-di-
rection. The normal velocity is assumed to be zero over the remainder of the plane.
This type of model is called a rigid baffle model since the element is assumed to be
embedded in an otherwise motionless plane, as discussed in Chap. 1. The motion of
the element radiates a 2-D pressure wave field p ( x, z , t ) into an ideal compressible
fluid medium that occupies the region z ≥ 0.

L. W. Schmerr Jr., Fundamentals of Ultrasonic Phased Arrays, 17


Solid Mechanics and Its Applications 215, DOI 10.1007/978-3-319-07272-2_2,
© Springer International Publishing Switzerland 2015
18 2  Acoustic Field of a 1-D Array Element

Fig. 2.1   Model of a 1-D


element radiating into a fluid
with density and wave speed,
(ρ, c) , respectively

As shown in many texts (see [Schmerr], for example) the application of New-
( )
ton’s law ∑ F = ma to a small fluid element yields the equations of motion (for
no body forces) of the fluid given by:

(2.1) ∂2u
−∇p = ρ 2 ,
∂t

Where ρ is the density of the fluid, u( x, z , t ) is the vector displacement, and the
2-D gradient operator ∇ = e x ∂ + e z ∂ , and (e x , e z ) are unit vectors in the x- and
∂x ∂z
z-directions, respectively. For an ideal compressible fluid the pressure in the fluid is
related to the fluid motion by the constitutive equation

 p = − λ B ∇· u, (2.2)

where λ B is the bulk modulus of the fluid. The quantity ∇·u appearing in Eq. (2.2)
is called the dilatation. Physically, it represents the relative change of volume per
unit volume of a small fluid element and it is also called the volumetric strain of the
fluid element [Schmerr]. The minus sign is present in Eq. (2.2) because a positive
pressure causes a decrease in the volume of a compressible fluid.
If one takes the divergence ( ∇·) of both sides of Eq. (2.1) and uses Eq. (2.2), it
follows that the pressure p ( x, z , t ) must satisfy the wave equation:

∂2 p ∂2 p 1 ∂2 p (2.3)
+ − = 0,
∂x 2 ∂z 2 c 2 ∂t 2
2.1  Single Element Transducer Models (2-D) 19

where the wave speed, c, in the fluid is given by

(2.4)
c = λB ρ.

In modeling waves in the fluid, we will assume that all the waves have a harmonic
time dependency exp( −iω t ) so that
(2.5)
p ( x, z , t ) = p ( x, z , ω ) exp( −iω t ).

Placing this relationship into Eq. (2.3) shows that p ( x, z , ω ) must satisfy the Helm-
holtz equation
(2.6) ∂ 2 p ∂ 2 p ω 2
+ + p = 0.
∂x 2 ∂z 2 c 2

Alternatively, we can view a solution p ( x, z , ω ) of Eq. (2.6) as the Fourier trans-


form (frequency domain spectrum) of a time dependent wave field p ( x, z , t ), where
 +∞
(2.7)
p ( x, z , ω ) = ∫
−∞
p ( x, z , t ) exp(+iω t )dt

and
+∞
(2.8) 1
p ( x, z , t ) = ∫ p ( x, z, ω ) exp(−iωt )dω ,
2π −∞

since if we take the Fourier transform of the wave equation it follows that the trans-
formed pressure p ( x, z , ω ) also must satisfy the Helmholtz equation.
We will solve our models of transducer behavior in this and later Chapters for
the fields p ( x, z , ω ). Since we will be working almost exclusively with frequency
domain wave fields in this book, we will henceforth drop the tilde on our frequen-
cy domain variables and simply write fields such as the pressure or velocity as
p ( x, z , ω ) or v ( x, z , ω ), etc. with the understanding that an additional time depen-
dent term exp( −iω t ) is also always present implicitly if we want to recover a time
domain solution (see Eq. (2.8)) or if we consider these fields as harmonic wave
fields.
To solve for the waves generated in the geometry of Fig. 2.1, we first note that
the Helmholtz equation has harmonic wave solutions given by
 p = exp(ik x x + ik z z ), (2.9)

where
 k 2 − k 2 k ≥ k
(2.10)
kz = 
x x

i k x − k k < kx
2 2

You might also like